Search Results

CalprestNG is a quantitative ELISA for detecting concentration of fecal calprotectin. CalprestNG can in vitro diagnostic to aid in the diagnosis of Inflammatory Bowel Diseases (IBD, Crohn's disease and ulcerative colitis) and to differentiate IBD from Irritable Bowel Syndrome (IBS) in conjunction with other clinical and laboratory findings.

Calprest®NG is an enzyme-linked immunosorbent assay (ELISA) system with colorimetric detection based on the use of antibodies against calprotectin. Calprotectin present in the diluted sample is bound by the antibody adsorbed to the surface of the plastic well. The enzyme conjugated antibody binds to the captured antigen and subsequently the enzyme catalyzes the conversion of the substrate to a colored product. The intensity of the color is proportional to the amount of conjugate bound, and thus to the amount of captured calprotectin. Concentration of calprotectin in the samples is calculated using the provided standards.

EasyCal is a single-use device for stool sample pre-analytical processing that allows the extraction of calprotectin from the specific amount of collected fecal sample required to perform Eurospital's Calprest® and Calprest®NG assays.

The device consists of a tube, containing 2.8 ml of extraction solution, a stick shaped with seven grooves for collecting the sample. The upper end is made up by two components which can be removed by opposite rotations. The screw cap (white) connected to the shaped stick traps the sample excess and can be then removed by counter-clockwise rotation. Once the extraction procedure has been completed, the sample can be transferred to an automated ELISA instrumentation, placing it directly into the sample rack. EasyCal allows an easy, reliable and reproducible way to sample from primary containers and analyze the extract directly from the device, without the need to weight the stool sample.

The document describes the CalprestNG device, an ELISA (enzyme-linked immunosorbent assay) for detecting fecal calprotectin, and its new pre-analytical processing accessory, EasyCal. The submission is for K191592, establishing substantial equivalence to the predicate device CalprestNG (K160447), which used manual extraction. The core of the study is to demonstrate that using EasyCal for sample extraction yields comparable results to the manual extraction method.

Here's an analysis of the acceptance criteria and the study proving the device meets them:

1. A table of acceptance criteria and the reported device performance:

The document primarily focuses on demonstrating the comparability of CalprestNG with EasyCal to CalprestNG with manual extraction, rather than predefined quantitative acceptance criteria with specific ranges. Instead, the "acceptance criteria" are implied by showing the comparability of the new method to the predicate method.

However, we can infer the performance metrics tested and their results, which serve as evidence of acceptance.

Table 1. Method Comparison and Performance

• Slope (95% CI): 0.9778 (0.901 to 1.003)
• Y-intercept (95% CI): 0.8495 (-0.8360 to 3.978)
• Correlation (r): 0.954
• Bias at 120 mg/kg (95% CI): -1.5% (-9.0% to 1.5%) |
  | Qualitative Agreement | High agreement (Negative, Positive, and Total Agreement) when classifying samples based on a clinical cut-off. | Referencing 120 mg/kg cut-off (borderline as positive/negative based on interpretation):
• Borderline as positive:
  • Negative Agreement (95% CI): 96.2% (81.1 to 99.3%)
  • Positive Agreement (95% CI): 100.0% (95.1 to 100%)
  • Total Agreement (95% CI): 99.0% (94.6 to 99.8%)
• Borderline as negative:
  • Negative Agreement (95% CI): 100.0% (90.1 to 100%)
  • Positive Agreement (95% CI): 95.4% (87.3 to 98.4%)
  • Total Agreement (95% CI): 97.0% (91.5 to 99.0%) |
    | Stool Sample Collection Performance| Maintain consistent and appropriate sample weight. | Mean weight collected by EasyCal: 56 mg. (This is a specific internal metric, the acceptance criteria implicitly being that this weight is consistent and adequate for the assay). |
    | Extraction Reproducibility | Low coefficient of variation (CV%) for within-run, between-day, within-operator, between-operator, and within-laboratory precision.| CV% for 8 samples across quantification range (average or example values presented): Representative CVs range from 3.8% to 6.9% for repeatability, and overall within-laboratory CVs up to 18.3%. (Specific ranges were not given as 'criteria', but the detailed table data demonstrates the measured precision). The study concludes: "Reproducibility of results obtained with Calprest® with EasyCal is confirmed." |
    | Extracted Sample Stability | Extracted samples should remain stable at specific temperatures and after freeze/thaw cycles for defined durations. | Samples met acceptance criteria for:
• 2-8 °C up to 21 days (recommendation: up to 14 days)
• Room temperature up to 73 hours (recommendation: up to 72 hours)
• Up to 5 freeze/thaw cycles (recommendation: not to exceed 4 cycles) |
  | EasyCal Device Stability (Shelf Life)| pH and volume stability over time; no adverse effect on CalprestNG performance. | pH and Volume: Between 7.77 and 7.84 (mean 7.80) for pH; 99.4% to 100.6% (mean 100.0%) for volume variation up to 25 months. All met acceptance criteria.
  CalprestNG performance: All acceptance criteria met at 25 months. Device stable for 24 months at 4 °C and does not affect CalprestNG performance. |
  | EasyCal Device Stability (Room Temp.)| Maintain performance when stored at room temperature for specified duration. | All acceptance criteria were met. Device stable for 72 hours at room temperature and doesn't affect CalprestNG performances. |

2. Sample sizes used for the test set and the data provenance

• Method Comparison (Stool extraction method comparison: EasyCal vs manual extraction procedure):

  • Sample Size: One hundred (100) stool samples.
  • Data Provenance: Not explicitly stated, but clinical studies for such devices typically involve prospective collection or use of banked samples from clinical sites. The document implies these are "patient samples" (from CLSI EP09c reference). Country of origin is not specified.
  • Retrospective/Prospective: Not explicitly stated for this particular sample set, but device validation studies often use a mix. Given the "CLSI EP09c" reference, it points to a formal comparison study, which could be prospective collection for the purpose of the study.

• Reproducibility study: Extraction Reproducibility:

  • Sample Size: Eight (8) stool samples (tested in replicate of five, once a day, for five days, by three different operators). This results in 8 samples * 5 replicates * 5 days * 3 operators = 600 total test points.
  • Data Provenance: Not specified.
  • Retrospective/Prospective: Not specified.

• Stool sample collection performance of EasyCal:

  • Sample Size: Five (5) different human stool samples. Each collected in replicates of five by three independent operators (5 samples * 5 replicates * 3 operators = 75 total weight measurements).
  • Data Provenance: Not specified.
  • Retrospective/Prospective: Not specified.

• Samples stability and handling:

  • Sample Size: Eight (8) stool samples.
  • Data Provenance: Not specified.
  • Retrospective/Prospective: Not specified.

• EasyCal device stability (Shelf life and Room Temperature):

  • Sample Size: Three lots of EasyCal devices were used for pH/volume tests. Twelve (12) samples were extracted for performance validation with aged vs. fresh devices. Eight (8) stool samples were used for room temperature storage validation.
  • Data Provenance: Not specified.
  • Retrospective/Prospective: Real-time stability studies are inherently prospective observations over time.

3. Number of experts used to establish the ground truth for the test set and the qualifications of those experts (e.g. radiologist with 10 years of experience)

This device is an in vitro diagnostic (IVD) quantitative assay for fecal calprotectin. The "ground truth" for such devices is established through:

• Referenced methods: The manual extraction method of the predicate device (K160447) served as the reference or comparative "truth" for the EasyCal method. The study aimed to show equivalence to this established method.
• Known concentrations: For reproducibility and stability studies, often spiked samples or samples with predetermined concentrations are used, verified by validated analytical methods.
• Clinical correlation (indirect): The output of the assay (Calprotectin concentration) is correlated with clinical findings (IBD vs. IBS diagnosis). However, the direct clinical diagnostic "ground truth" (e.g., patient biopsy, endoscopy results) is not used within this specific analytical validation section to establish the "truth" for the quantitative assay itself. The clinical performance characteristics are stated to be "Same as approved in 510(k) submission # K160447," implying that clinical utility was established previously, and this submission focuses on analytical equivalence.

Therefore, human experts in the context of "ground truth establishment" for an image analysis or diagnostic interpretation task (like radiologists for imaging) are not directly applicable here. The "ground truth" is the precise measurement of calprotectin or the established manual extraction benchmark.

4. Adjudication method (e.g. 2+1, 3+1, none) for the test set

Adjudication methods like 2+1 or 3+1 are typically used in studies where human readers/experts independently interpret complex data (e.g., medical images), and a consensus or adjudication process is needed to resolve discrepancies and establish a robust ground truth for classification tasks.

This study is an analytical performance study of a quantitative in vitro diagnostic (IVD) device. The "truth" is either the measured concentration by a reference method or the established manual extraction procedure. Therefore, no adjudication method involving multiple human readers/experts is pertinent or mentioned for establishing the "ground truth" of the test set in this context.

5. If a multi reader multi case (MRMC) comparative effectiveness study was done, If so, what was the effect size of how much human readers improve with AI vs without AI assistance

No MRMC comparative effectiveness study was done or is relevant here. This is a submission for an IVD device measuring a biomarker (fecal calprotectin), not an AI-assisted diagnostic imaging or interpretation tool for human readers. There are no "human readers" interpreting the output in a way that would be "improved with AI assistance." The technology is a laboratory assay.

6. If a standalone (i.e. algorithm only without human-in-the-loop performance) was done

This is an analytical device, and its performance is inherently "standalone" in terms of its measurement capabilities. The CalprestNG ELISA, whether with manual or EasyCal extraction, provides a quantitative result for fecal calprotectin. There isn't an "algorithm" in the sense of AI or machine learning that processes images or complex data for interpretation. The "algorithm" is the biochemical assay itself, which delivers a numerical value.

The performance studies (method comparison, reproducibility) directly evaluate this standalone performance. The "human-in-the-loop" aspect relates to standard laboratory practices (performing the assay, pipetting, reading results from the instrument), but not an assistive AI.

7. The type of ground truth used (expert consensus, pathology, outcomes data, etc)

The "ground truth" for the test set in this analytical validation is established by:

• Reference Method Comparison: The results obtained via the CalprestNG assay with the manual extraction procedure (K160447 predicate device) served as the primary reference method to which the new EasyCal extraction method was compared for equivalence.
• Known samples/standards: For reproducibility and stability, samples with known or well-characterized concentrations of calprotectin (likely established using a highly accurate and precise method) are typically used.
• Internal standards and controls: The assay uses calibrators (0, 2.5, 12.5, 25, 50, 150 ng/ml) to quantify calprotectin concentration, which act as internal "ground truth" for the assay's operational range.

8. The sample size for the training set

This document describes the validation of an in vitro diagnostic assay kit, not a machine learning or AI model that requires a "training set" to learn from data. Therefore, the concept of a "training set" in the AI/ML sense is not applicable here. The assay is based on established biochemical principles (ELISA) and does not "learn" from data in the way an algorithm does.

9. How the ground truth for the training set was established

As explained above, there is no "training set" in the context of AI/ML for this device. The "ground truth" for the assay's core function is inherent in its biochemical design, the use of validated reagents, and the calibration curves established with known standards. The validation studies described in the document (method comparison, reproducibility, stability) assess the performance of the final manufactured product against predefined analytical metrics and comparison to a predicate device.

Ask a specific question about this device